JP2007115162A - Double storage method of data - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To ensure reconfiguration when abnormality occurs in one hard disk device in a mirror disk device. <P>SOLUTION: A computer body operates a program for issuing a command for performing reading from respective hard disk devices 2A and 2B in a background, makes the mirror disk device check data match/mismatch of the hard disk devices by this reading, and makes, when mismatching or reading error, the mirror disk device repair a defective block by overwriting the relevant block of the hard disk device with abnormality using the relevant block data of the sound hard disk device. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to a data double storage method for enhancing the reliability of data by storing the same data in two hard disk devices.

  This type of double data storage is implemented by data storage in an industrial computer that requires high reliability, and is called a mirror disk device (RAID 1). As shown in FIG. 2, this mirror disk device appears to be one hard disk device from the computer main body 1 side, but the same data is written to the two hard disk devices 2A and 2B on the hard disk device side.

This mirror disk device reduces the risk of data loss due to a failure of one of the hard disk devices without changing the driver software etc. on the computer main body side, and improves reliability (for example, see Patent Document 1).
Japanese Patent Laid-Open No. 10-066047

  In the hard disk device, although rare, there are cases where the written data becomes abnormal due to an ECC error (a phenomenon in which part of the data is broken for some reason after writing), an illegal block, or the like. Since the capacity of recent hard disk devices has become extremely large, the probability that this abnormal part will occur increases. This abnormal part can be detected as a data abnormality by performing a reading process from the hard disk device.

  In the processing performed in the mirror disk device, as shown in FIG. 3A, the same data is written into the two hard disk devices 2A and 2B when a write request is made by software on the computer main body side. On the other hand, as shown in FIG. 3B, when a read request is made by the computer main body side, the data is read from a hard disk device on one side, for example, 2A, from which the corresponding data can be read quickly.

  Here, as shown in FIG. 4A, when the above abnormality (reading error) is detected when reading data from the hard disk device 2A, as shown in FIG. 4B, the other side of the mirror disk device is detected. The corresponding data is read from the hard disk device 2B, and the data at the abnormal location is overwritten (or replaced block processing) to restore the data and ensure its correctness.

  However, when data is read infrequently or when data is read from a hard disk device that is not abnormal, or when there is data in an area that is not used as a file system, the above error has occurred for these data. In some cases, it is difficult to detect in normal operation, and the repair function may not work.

  If the one-side disk device 2B fails in this state, as shown in FIG. 5 (a), the mirror disk device makes the two hard disk devices 2A and 2B the same data. All data is copied from the device 2A to the replaced hard disk device 2B. At this time, as shown in FIG. 5B, if there is an abnormal portion of data in the hard disk device (copy source) 2A on the non-replacement side, an error occurs when reading the copy source, and as a result normal In addition, the mirror disk device cannot be constructed.

  In addition, since an abnormal location results in data loss, it is necessary to take measures such as reformatting the hard disk device or restoring from a backup.

  An object of the present invention is to provide a data double storage method that solves the above-described problems.

  In order to solve the above-described problem, the present invention operates a program that issues a command for reading back from each hard disk device of the mirror disk device in the background, and there is data match / mismatch of each hard disk device by this read back. If a discrepancy occurs, the corresponding block data of the hard disk device on the healthy side is used to overwrite the corresponding block of the hard disk device where the error occurred, so that the defective block can be repaired. It is characterized by the following method.

(1) A mirror disk device composed of two hard disk devices stores data requested to be written from the computer main body in two hard disk devices, and requests data read from the computer main body from one hard disk device. In the data double storage method with a processing function that reads data, reads the corresponding data from the other hard disk device when an error is detected in this data reading, and overwrites or replaces the data at the abnormal location.
The computer main body has a process of periodically executing a command for causing reading simultaneously from each hard disk device of the mirror disk device in a background operation,
When the mirror disk device receives the command from the computer main body, the mirror disk device performs a process of reading back each of the hard disk devices, and the data of all the blocks read back in this process for each hard disk device. The process of performing a check for coincidence / non-coincidence with the error and the process of repairing the abnormal location by overwriting the read-back data of the hard disk device on the sound side in the error occurrence location of the hard disk device when a read-back error occurs in this inspection It is characterized by having.

  As described above, according to the present invention, a program for issuing a command for performing read back from each hard disk device of the mirror disk device is operated in the background, and whether or not there is a data match / mismatch of each hard disk device by this read back. If a mismatch occurs, the corresponding block data of the hard disk device on the healthy side is overwritten on the corresponding block of the hard disk device in which an abnormality has occurred, and the following effects are obtained.

  (1) When the mirror disk device is replaced due to a failure of the one-side hard disk device, it is possible to prevent a mirror disk device reconstruction error due to an abnormal block on the hard disk device side that has not failed.

  (2) It is possible to eliminate the system stop due to the system recovery work (restoration work from the backup medium) due to the abnormality, and the system reliability can be improved.

  FIG. 1 shows an embodiment of the present invention, and is a processing procedure when a mirror disk device is replaced due to a failure of one hard disk device.

  First, a program for periodically transmitting a command (referred to as a VERIFY command) that causes simultaneous reading from each hard disk device 2A, 2B of the mirror disk device is prepared on the computer body side. This VERIFY command is a SCSI (small computer system interface) protocol in a standard hard disk device, and is a command for performing a read back process of a specified logical block in the hard disk device.

  In the present embodiment, the computer main body periodically executes the above VERIFY command as a background operation. In response to this VERIFY command, as shown in FIG. 1A, the mirror disk device performs a read-back process on each of the two hard disk devices 2A and 2B when receiving the VERIFY command. For all blocks of data read back from each hard disk device 2A, 2B, as shown in FIG. 1 (b), the mirror disk device performs a data match / mismatch check for each hard disk device, When a read-back error occurs in this inspection (in the figure, a read-back error occurs in the data on the hard disk device 2A side), as shown in FIG. The read-back data of the hard disk device 2B is overwritten (copied with the corresponding data) at the location where the abnormality occurred, and the abnormal location is repaired.

  If an error occurs in the hard disk device 2B by reading back with the VERIFY command, it is overwritten with the data of the corresponding block from the hard disk device 2A.

  Therefore, in the mirror disk device, when replacement due to failure of one hard disk device occurs, it is possible to prevent the mirror disk device reconstruction abnormality due to the abnormal block to the healthy hard disk device that has not failed and to repair the abnormal part .

  In the above description, the case where the bus of the computer is SCSI and the VERIFY command is executed has been described. However, the present invention is not limited to SCSI. Other than this, for example, an IDE or USB bus can perform the same processing.

The disk rebuilding process sequence diagram which shows the embodiment of the present invention. The block diagram of a mirror disk apparatus. The read / write processing procedure diagram of the mirror disk device (conventional). The processing procedure figure at the time of abnormality detection (conventional). Disk rebuild processing procedure diagram (conventional).

Explanation of symbols

1 Computer body 2A, 2B Mirror disk device

Claims (1)

A mirror disk device composed of two hard disk devices stores data requested to be written from the computer main body in two hard disk devices, and reads data from one hard disk device in response to a data read request from the computer main body. In this case, when an error is detected in the data reading, the corresponding data is read from the other hard disk device, and the data is stored in the double storage method having a processing function for performing overwriting or alternation block processing on the abnormal portion data.
The computer main body has a process of periodically executing a command for causing reading simultaneously from each hard disk device of the mirror disk device in a background operation,
When the mirror disk device receives the command from the computer main body, the mirror disk device performs a process of reading back each of the hard disk devices, and the data of all the blocks read back in this process for each hard disk device. The process of performing a check for coincidence / non-coincidence with the error and the process of repairing the abnormal location by overwriting the read-back data of the hard disk device on the sound side in the error occurrence location of the hard disk device when a read-back error occurs in this inspection And a method for double storage of data.

Images